In this study, we found pieces of robust evidence about causal association between child and adult BMI and the risk of infertility and cervical cancers. In details, the higher genetic score of adult and childhood BMI was associated with increased of PCOS. Consistent with our conclusions, observational studies and meta-analyses have uncovered the same outcome that women with PCOS had higher risk of central obesity(Lim et al., 2012) and the prevalence of PCOS was increased among overweight and obese women(Alvarez-Blasco et al., 2006), earlier adiposity rebound and severe BMI rise in childhood predicted later development of PCOS(Koivuaho et al., 2019b). The negative causality between child BMI and endometriosis through IVW method was convincing because observational studies has come up with same conclusion that there was an inverse relation between body size at 5,10,20 age and the disease(Vitonis et al., 2010). The prospective case-control aiming at more than 100,000 women from the United States for over 20 years exhibited the inverse tendency to develop endometriosis was more significantly obvious in 18-year-old BMI than current BMI, persisted noteworthy in both infertile women and those without concurrent infertile. An genome-wide enrichment analysis demonstrated a significant enrichment of common variants overlapping both endometriosis and waist-to-hip ratio adjusted for BMI, representing for fat distribution(Rahmioglu et al., 2015). There is another case-control analysis in Australia revealed that women who self-reported overweight at age 10 had an increasing risk of endometriosis(Nagle et al., 2009). However, mother-daughter pair reports came up an opposite association that underweight at 16 years old positively related to endometriosis, which consisted with ours. Nevertheless, we didn’t acquire same positive findings about BMI associated with endometrial cancers, ovarian cancers using the method as other MR analysis and observation did. A MR analysis published in 2016 which uses 4 subsets of cases and controls from EC datasets of Australian and European ancestry(Painter et al., 2016). The final IVW result was combined using random effects meta-analysis after stratified into quartiles and calculated separately. So different races and computing method may contribute to the divergence in our results. While using the same adult BMI-SNPs without abandoning 2 loci in strong linkage disequilibrium and different outcome database, Gao, C., et al.(Gao et al., 2016) concluded that 1 standard deviation in genetically predicted adult BMI was associated with 35% increase risk in overall ovarian cancer. However, after excluding overlap loci, the significance disappeared. They also didn’t find strong association between genetically predicted child BMI with ovarian cancer risk. Another MR study ended up with a consequence that BMI as instrumental invariables as obesity, increased risk of non-high grade serous ovarian cancer of European ancestry, but had nothing to do with high grade serous ovarian cancer(Dixon et al., 2016). This reminds us that subtypes of ovarian cancer may react to obesity differently and contribute together to nonsense of overall OC to obese. The inverse causal relationship between cervical cancers and adult BMI need to be confirmed because the p value of IVW reached 6.57*10 − 2 while WM concluded significantly. A retrospective cohort study of patients of cervical cancer(Frumovitz et al., 2014) were classified as underweight (BMI < 18.5kg/m2), normal weight (BMI 18.5-24.9kg/m2), overweight (BMI 25-29.9kg/m2), obese (BMI 30-34.9 kg/m2) and morbidly obese (BMI ≥ 35.0kg/m2). After controlling for prognostic factors, only morbidly obese remained an independent risk for mortality of CC. So, the classification left us a hint that the extent of obesity or staging of disease could obscure the real causal association of risk factors and results.
To the best of our knowledge, it is the first mendelian randomization analysis of the causality of BMI and PCOS or endometriosis. Also, never ever had articles published the association between childhood BMI and common gynecological cancers. With the appearance of genome wide associated study, we can take advantage of information sharing to further MR research on causality of exposures and diseases. Previously, researchers get used to carrying out retrospective studies or prospective studies. Compared to the two classical research methods, strengths of our MR analysis can be listed as follow: 1) less vulnerable to reverse causation or confounding bias from economic and educational levels or lifestyle like smoking or alcoholism; 2) insusceptible of experimental such as acting time and degree of disposal; 3) the sample capacity can be big enough as soon as getting access to database; 4) our results show besides adult obesity, severe child obese can already be predictable to endometriosis and PCOS in later life, so it’s vital for us to prevent the trend of obesity in childhood and take measures to keep in balanced shape whenever weight gains, and if obese has taken place, there’s a need for to lose weight; 5) MR can directly explains the causality between exposure and outcomes not only simple correlation. So, drug use and side-effect of weight gain need to be notice for future health. Of course, there is room for improvement: 1) BMI is not a perfect index for obesity, as we can see, waist to hip ratio as well as subcutaneous fat thickness is often chosen to be proxies of obesity. Fat distribution rather than BMI can also be an independent risk of diseases; 2) different races can have changes in instrumental variables and genetic score of outcome; Here in our MR analysis, results only applied to European; 3) there may be some overlapped loci in adult and child BMI associated SNPs, there’s a possibility that genetic variants influence both of them; 4) we can’t count out the probability of population stratification and other potential confounding, severity levels of outcomes and stages of diseases should be underlined and distinguished if possible, and our final association was not obvious.